7KH(YROXWLRQRI(SLJHQHWLFV *DU\)HOVHQIHOG
Perspectives in Biology and Medicine, Volume 57, Number 1, Winter 2014, pp. 132-148 (Article) 3XEOLVKHGE\7KH-RKQV+RSNLQV8QLYHUVLW\3UHVV DOI: 10.1353/pbm.2014.0004
For additional information about this article http://muse.jhu.edu/journals/pbm/summary/v057/57.1.felsenfeld.html
Access provided by your local institution (24 Apr 2015 05:20 GMT)
The Evolution of Epigenetics
Gary Felsenfeld
ABSTRACT Early studies of the developing embryo raised the question of how
a fertilized egg could give rise to a complex multicellular organism containing many different kinds of cells. The term epigenetics originally referred to the study of these processes. With the advent of detailed knowledge of mechanisms of gene expression, this definition was superseded by another: epigenetics concerned the transmission of phenotype through mitosis or the germ line by mechanisms that did not involve changes in the DNA sequence. Much effort has been spent in attempting to identify and characterize these events. Work initially focused on DNA methylation as an epigenetic mark, but more recently there has been an emphasis on histone modifications as possible carriers of epigenetic information. However, there is confusion between situations in which the modifications may be propagated through cell division, thus helping to maintain a pattern of gene expression, and situations in which the modifications are simply part of the transcriptional apparatus. Arguments about the role of the histones have led to a reexamination of the definition of epigenetics and the primary events in development leading to cell type specific gene expression patterns.
S
ince the early days of embryology, a central puzzle for biologists has been how a fertilized egg can execute a clearly defined and reproducible program that leads ultimately to a complex organism. It was clear that all of the information
Laboratory of Molecular Biology, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, MD 20892. Email: [email protected]. This work was supported by the Intramural Research Program of the National Institute of Diabetes and Digestive and Kidney Diseases, NIH. Perspectives in Biology and Medicine, volume 57, number 1 (winter 2014): 132–148. © 2014 by Johns Hopkins University Press
132
The Evolution of Epigenetics
necessary to create the adult must already reside in the zygote, but how that information was translated into a complex organism was obscure. Even as recently as the late 1940s, the molecular mechanisms associated with early development were unknown and, in a sense, unknowable, given the available knowledge and technology. In the preceding centuries, embryologists had described in detail the progression of events during embryonic development, in both vertebrates and invertebrates. Two kinds of developmental models were considered. One assumed that the zygote already contained a preformed structure that simply expanded; the other proposed that development involved a preprogrammed series of chemical reactions required to generate the observed complexity. The discovery of chromosomes by Flemming in 1879 and the accumulating evidence that they carried the genetic information seemed to eliminate the first model. (For a complete description of the state of knowledge in 1900, see Wilson 1900.) In 1911, Morgan reported experiments in Drosophila that led him to two principal conclusions: “First, that sex-limited inheritance is explicable on the assumption that one of the material factors of a sex-limited character is carried by the same chromosomes that carry the material factor for femaleness. Second, that the ‘association’ of certain characters in inheritance is due to the proximity in the chromosomes of the chemical substances (factors) that are essential for the production of those characters” (365). Although it became accepted that the genetic information specifying the organism must reside in the chromosomes, and although individual “characters” could be associated with specific chromosomal loci, the chemical nature of the storage mechanism was unknown, as was the connection between the genes and the developmental process. At the same time, embryologists were cataloging the events of development using a different vocabulary and point of view. As described in a recent account of this period, embryologists were not especially interested in genes, because they were seen as static carriers of genetic information and therefore not capable of giving rise to a developmental program that varied among cell types (Gilbert 2012). Embryology and genetics were seen as distinct disciplines, addressing different questions about distinct mechanisms. In publications beginning in the late 1930s, embryologist C. H. Waddington attempted to make clear the connection between genes and developmental processes.The term epigenetics was used first by Waddington in 1942 to describe the study of the “causal mechanisms” by which “the genes of the genotype bring about phenotypic effects” (18). He saw clearly that specific patterns of expression of multiple genes were associated with each developmental stage and cell type. Waddington’s definition of epigenetics (which will be called D1) thus corresponded more or less to our contemporary view of the developmental process. All of these definitions were necessarily imprecise, because the roles of the many components of the chromosomes, as well as the other contents of the nucleus and cytoplasm, were not clear. Early work of A. H. Sturtevant (1925) and H. J. Muller (1930) had shown, however, that certain chromosome translocations in Drosophila
winter 2014 • volume 57, number 1
133
Gary Felsenfeld
could result in phenotypic changes even though there was no observable change in the chromosome structure within the displaced region. Many subsequent studies of such position effects made it clear that, whatever the molecular structure and mode of action of the individual “genes,” they did not function independently of a more extended environment. During this period, there was continuing debate about the relative contributions of the nucleus and cytoplasm to heredity (for a detailed description of these varied points of view, see review by Haig 2004). This debate extended not only beyond the 1944 experiments of Avery and colleagues, which provided strong evidence that DNA was the genetic material, but also beyond 1953, when the revelation of the DNA structure made clear where and how the genetic information might be stored.There was, in fact, continued doubt that DNA alone, free of proteins or other cellular components, could carry out this function. The idea that somatic cell nuclei contained the full complement of DNA was not confirmed until the experiments of Briggs and King (1952) and the definitive study by Laskey and Gurdon (1970), which showed that a nucleus from a somatic cell, introduced into an enucleated oocyte, could give rise to a normal adult. The problem of development, starting with the zygote, seemed to be reduced to a question of how individual genes were controlled in different cell types. Phenomena such as position effect variegation and X chromosome inactivation indicated, however, that heritable phenotypes might involve something in addition to DNA sequence. Early studies of gene regulation in bacteria and bacterial viruses suggested ways in which cellular expression states could be preserved through cell division. A trivial example would be a system in which a gene product was required for the activation of its own gene: provided that dilution during cell division did not reduce the abundance of the protein below some critical level, the active gene state would be maintained indefinitely through cell division. A more elaborate proposal from Delbrück involved a hypothetical pair of pathways, each producing an inhibitor of the other, and giving rise to a bi-stable expression system. An actual example of such a mechanism is provided by the switch between the lytic and lysogenic states in lambda bacteriophage, in which transient expression of a protein is sufficient to trigger a switch in states that is maintained through rounds of bacterial cell division (Ptashne 1992; Ptashne, Johnson, and Pabo 1982). In 1975, Riggs and Holliday and Pugh proposed that DNA methylation might in principle serve as a stable mark that could affect binding of transcription factors, hence gene expression, and that also could be propagated through cell division, if—as clearly stated by Riggs—there existed an enzyme that could methylate preferentially hemi-methylated sites, so that a palindromic site methylated on one strand would automatically be methylated on the second strand following replication (Figure 1). The existence of such systems was subsequently demonstrated by Bird (1978), and work in many laboratories has demonstrated the important role of DNA methylation in regulation of gene expression in vertebrates (Razin and Cedar 1977; Stein et al. 1982). 134
Perspectives in Biology and Medicine
The Evolution of Epigenetics
Figure 1 Propagation of DNA methylation patterns during DNA replication. A typical MeCpG site is modified on both strands. Following replication, the hemi-methylated site is recognized by the methyltransferase Dnmt1, which methylates the CpG on the newly synthesized strand.
The persistent debate about the relative contribution of mechanisms of cellular inheritance not connected directly to DNA sequence, probably reinforced by the increasingly detailed understanding of gene regulatory mechanisms, ultimately led to a change of focus in the way that the word epigenetics was used. The new definition (definition D2) became “the study of mitotically or meiotically heritable changes in gene function that cannot be explained by changes in DNA sequence.” To the extent that this might be viewed as a refinement of Waddington’s definition, it carries the unfortunate connotation that developmental processes should be ascribed principally to epigenetic mechanisms as newly defined—a point of view that has led to some confusion, as discussed below. DNA methylation was not the only phenomenon that could fit the new definition of epigenetics. In Drosophila, which lacks significant DNA methylation, the Polycomb protein complex was shown to shut down “permanently” genes no longer required in development, a state transmitted through cell division (for a recent review, see Simon and Kingston 2013). Specific X chromosome inactivation in vertebrate females appeared to be transmitted to daughter chromosomes, and at imprinted loci such as the Igf2/H19 region, only one of the two alleles was expressed, in a parent-of-origin–specific manner maintained through cell division. DNA methylation was implicated at least partially in these phenomena, but there winter 2014 • volume 57, number 1
135
Gary Felsenfeld
were other, major differences between the structures associated with the active and the inactive chromosome or alleles.
Epigenetics as Chromatin Structure The earliest experimental information about how gene expression was regulated came from studies in bacteria and bacteriophage. Jacob and Monod (1961) characterized the lac operon in E. coli; by the mid-1960s, the Gilbert laboratory had identified specific DNA-binding lac repressor protein (Gilbert and MüllerHill 1966), and Ptashne (1967) had isolated the lambda phage repressor.The model of transcriptional repression (and later, activation) through binding of regulatory proteins near the transcription start site, at “promoters,” was extended to eukaryotes, where it became clear that more distant sites, “enhancers,” also contributed to regulation of expression. It had long been obvious, however, that the organization of the genetic material in eukaryotes was much more elaborate: within the nucleus, the DNA is packaged as chromatin, a complex of DNA with proteins, principally the positively charged histones.These were at first envisioned as functioning to compact the large amounts of genomic DNA so that they would fit into the nucleus, and they do serve that purpose. It was also obvious that the histones might prevent DNA from being transcribed or replicated, unless there were specific mechanisms to modify or displace them. Well before the work in bacteria described above, and before it was proven that all somatic cells contained the same complement of DNA, Stedman and Stedman (1950) proposed that the histones acted as transcriptional repressors to silence genes in a cell-type specific manner, which would explain how different cell types had different phenotypes—the question that embryologists had been asking for many years. That point of view has in part survived, with modifications, in many models of chromatin function. One role of chromatin function is certainly to reduce expression of silenced genes to levels much lower than can be observed in prokaryote systems, but this model does not explain where the information comes from to instruct the histones which genes to repress, so the fundamental question is not answered. It was another quarter-century before it was discovered that the histones organized DNA through the formation of repeating subunits, called nucleosomes (Kornberg and Thomas 1974). Each nucleosome contains an octamer of the “core” histones: two each of H2A, H2B, H3, and H4, with two superhelical turns, 165 base pairs of DNA, wrapped around the outside, providing the first level of DNA compaction inside the nucleus (Figure 2). A single molecule of a separate histone family (H1) is typically bound in vivo at the site where the DNA enters and exits the nucleosome. Most work, however, has been done with so-called nucleosome core particles, which have 147 base pairs of DNA wrapped in 1¾ turns, and lack H1. The fact that the core histones are highly conserved in evolution suggested that they must have functions besides DNA compaction. We now know, through work 136
Perspectives in Biology and Medicine
The Evolution of Epigenetics
Figure 2 High resolution structure of the nucleosome core particle. The alpha-helical domains of the four core histones (two copies each of H2A, H2B, H3, and H4) are visible. The histone “tails,” which extend from the particle, are accessible for modification. Source: Luger et al. 1997.
in many laboratories, that the core histones are targets of covalent modifications that modify their properties and can to a considerable extent be correlated with the function of the DNA sequences with which they are associated. Most but not all of these modifications are located on the amino terminal “tails” of the histones, which protrude from the nucleosome and are accessible to enzymes (see Figure 2). Some of the known modifications are shown in Figure 3. They include acetylation of lysines, methylation of lysines and arginines, and phosphorylation of serines and threonines. This is further complicated by the ability of lysine and arginine residues to be either mono-, di-, or (in the case of lysine) tri-methylated. Each modification typically requires a specific enzyme, but there may be more than one specific enzyme capable of that modification. Which one of these is used at a particular site in vivo may depend on the nature of the site. All of these modifications are reversible: again, specific enzymes are responsible for removing each kind of modification. Do these modifications affect function, and if so, is there a simple relationship between a particular modification and a corresponding function? Early experiments, especially in budding yeast (S. cerevisiae), showed that interfering with the function of individual histone modifying enzymes led to distinct phenotypes (Brownell et al. 1996; Taunton, Hassig, and Schreiber 1996). This has been confirmed by countless experiments in cells from higher eukaryotes. Thus, there can be little question that these histone marks have essential functions. But what are these functions? It
winter 2014 • volume 57, number 1
137
Gary Felsenfeld
Figure 3 Some of the chemical modifications found on the amino terminal tails of histones H3 and H4, and the associations often found between certain modifications and transcriptional activity.
was originally suggested that each modification would have a specific, identifiable effect on some aspect of DNA biochemistry, constituting a code that would allow correlating the modification with a particular DNA-dependent process: transcription initiation or elongation, enhancer or promoter activity, DNA repair or DNA replication. With the advent of ChIP-seq methods for mapping individual histone modifications genome-wide, it became possible to address these questions. It was evident that certain modifications had a high probability of being associated with particular regulatory features: for example, transcriptionally active genes are typically marked with acetylated histones.There are also strong correlations of certain histone H3 methylation marks with enhancers and promoters. Other methylation marks in higher eukaryotes are associated with silenced heterochromatic regions. Certain sites may display specific modification patterns, perhaps cell cycle dependent. There is, however, no simple set of rules that can account for all of the varied combinations of observed modifications; unlike the genetic code, these patterns cannot be decoded unambiguously. In budding yeast, for example, it has recently been shown that acetylated lysines on the N terminal tails of histones H2A and H4 have a redundant effect on phenotype (Kim et al. 2012). Despite the complexity of these relationships, it is clear that there are important correlations between histone modifications and gene activity, and that interfering with the modification pathways can have major effects in vivo. For that reason, investigators began referring to the modifications as “epigenetic marks,” perhaps by analogy to the DNA methylation mark. This has given rise to a heated debate, 138
Perspectives in Biology and Medicine
The Evolution of Epigenetics
discussed further below, that has resulted in attempts to further constrain the definition of epigenetics. But even without resorting to redefinition, the question remains of whether histone modifications, though clearly distinct from changes in DNA, can induce mitotically or meiotically heritable changes in gene function. To answer that question it would be helpful if we understood the biochemical relationship between histone modification and gene expression. Unfortunately, the answer appears to be getting more obscure as new data and points of view accumulate. Recent results, such as those of Kim et al. (2012), lend support to the idea that histone acetylation is generated as a consequence of transcription and serves to reduce the affinity of the histone octamer for DNA, making subsequent transcription events easier (Henikoff and Shilatifard 2011). Amusingly, this is an idea that was first advanced in the early studies of this modification, and there were in fact numerous reports showing that octamers containing acetylated histones are not bound as tightly to nucleosomal DNA. Even if one accepts this view that some modifications follow rather than precede transcription, it is important to remember that they do play an important role in transcriptional mechanisms; interference with the modification process has multiple effects on phenotype. Many proteins can recognize histone modifications as a signal to bind either more or less tightly to the nucleosome. For example, the protein HP1, often associated with silenced chromatin, binds to methylated lysine 9 of histone H3 (H3K9Me2/3), which in turn can lead to recruitment of the enzyme that carries out that methylation. Histone modifications can also lead to the recruitment of ATP-dependent nucleosome remodeling enzymes, capable of moving nucleosomes in such a way as to make specific DNA sequences more or less accessible. Some studies have documented an orderly series of events that occur during activation of specific genes (Im et al. 2005; Miranda et al. 2013). Furthermore, there can be a series of well-defined steps in which sites of DNA methylation can recruit enzymes that introduce “silencing” modifications, and in turn some sites of modification can recruit DNA methytransferase. Histone variants also contribute to regulation of gene expression. Among these, the H3 variant histone H3.3 confers reduced stability on nucleosomes in which it is incorporated; nucleosomes containing the variant H2A.Z in combination with H3.3 are even more unstable (Jin and Felsenfeld 2007). The H3 variant CENP-A is essential to centromere identity. Some of these variants have clear epigenetic functions, as discussed below.
Is Chromatin Structure Transmissible? To qualify as an epigenetic mechanism according to definition D2, a chromatin modification pattern or structure would have to be transmissible through cell division. The difficulty in addressing this question is in part related to the problem of distinguishing cause from effect. Can histone modifications, in the absence of other signals, create a specific active or silenced chromatin domain, or are they winter 2014 • volume 57, number 1
139
Gary Felsenfeld
always downstream consequences of earlier processes? It seems clear that during development the initial information to determine the activity state of a gene must come from sequence-specific transcription factors that recognize the regulatory elements associated with the gene. This is the only kind of interaction we know of that can convey that information (Figure 4). It follows that whatever the associated histone modifications, they must occur downstream of that initial event. If we are considering an individual cell, then the histone modification changes that follow binding of the transcription factor could most reasonably be viewed simply as biochemical processes that are part of the total regulatory apparatus: essential, but with no epigenetic connotations. As critics have often stated, it can be misleading to refer to histone modifications under these circumstances as “epigenetic marks.” The more relevant question, however, is whether histone marks, once established, can be transmitted through cell division or the germ line in such a way as to preserve the activity state of the associated genes without going through further de novo initiation by transcription factors as described above. In this model, chromatin structure, surviving through cell division, preserves the earlier-established “on” or “off ” state of the locus. Such behavior would qualify as epigenetic according to definition D2. One class of proposed mechanisms began with the observation that HP1 binds to H3K9Me2/3, and that HP1 in turn recruits the enzyme that carries out H3K9 methylation, creating in principle a system for propagation of the mark (Bannister et al. 2001; Jenuwein 2001). The functioning of such systems in vivo has been demonstrated most clearly for the Polycomb complex PRC2 and its associated modification, trimethylation of H3 lysine 27 (H3K27me3) (Hansen et al. 2008). PRC2 binds to H3K27me3, and binding to a methylated site helps activate the PRC2 complex member responsible for methylating H3K27, in principle allowing it to methylate a site on an adjacent nucleosome (Figure 5). A form of PRC2, in which one component was fused to GAL4, was targeted in cells to GAL4 binding sites upstream of a reporter gene integrated into the genome. This resulted in silencing of the reporter and increased H3K27 trimethylation. Most important, the methylation mark, which extended downstream of the target site for at least 1,600 base pairs, was maintained, due to the action of the endogenous PRC2 complex, over several population doublings after expression of GAL4-PRC2 was shut down. These and other results show how the Polycomb complex could propagate the K27 methylation mark along a chain of nucleosomes during interphase (Beisel and Paro 2011; Margueron and Reinberg 2010); the same mechanism could explain how the same signal could be maintained during and after cell division. Recent studies have shown that such processes are likely to be rather more complicated, perhaps involving another Polycomb complex, PRC1, or maintenance of Polycomb binding during DNA replication (see review by Simon and Kingston 2013). The evidence does seem to support some kind of epigenetic mechanism, but it is still not clear what biochemical processes, leading to silencing, are signaled by H3K27 trimethylation.
140
Perspectives in Biology and Medicine
The Evolution of Epigenetics
Figure 4 A simplified description of steps that may be associated with activation of gene transcription. The important point is that the initial step (top) must involve recognition of DNA sites (nucleotide sequences) near the gene by one or more specific regulatory proteins that target that gene. That is the signal that allows the remaining steps (below) to prepare the chromatin to accommodate RNA polymerase and transcription.
Histone variants have also been implicated in epigenetic processes. Because nucleosomes containing H3.3 are less stable than those containing the major variants and are found at transcriptionally active genes, it has been suggested that DNA covered with such H3.3 nucleosomes would be more readily transcribed (McKittrick et al. 2004; Ng and Gurdon 2008). Furthermore, nucleosomes containing H3.3 can be deposited throughout the cell cycle, whereas H3.1 or H3.2 are laid down only during S phase. Because the passage of RNA pol II tends to displace nucleosomes, the newly deposited ones will contain H3.3 and make transcription even easier. Most important, after replication the daughter gene copies will still contain H3.3 and tend to retain the “active” conformation, thus propagating that state. Experiments in which Xenopus somatic cell nuclei are transplanted into enucleated oocytes support a role for H3.3 in epigenetic transmission of phenotype (Ng and Gurdon 2008). A nucleus is taken from an embryonic endodermal cell, and the resulting embryo inappropriately expresses endoderm-specific genes in cells of the animal pole. If the experiment is repeated with a nucleus from one of these cells, the resulting embryo again has incorrectly programmed animal pole cells. These epigenetic effects are modulated by H3.3: overexpression of H3.3 increases and depletion decreases the number of inappropriately expressing cells.
winter 2014 • volume 57, number 1
141
Gary Felsenfeld
The most unambiguous example of an epigenetic process mediated solely by a histone is provided by the family of centromeric H3 variants, called CENP-A in vertebrates. Nucleosomes containing CENP-A are found only at centromeres, and they are essential to centromeric function during mitosis. It has been shown in Drosophila, and more recently in vertebrates, that the location of a centromere depends entirely on the presence of CENP-A. Although normal centromeres do contain special sequences, it is the presence of CENP-A that signals where centromeres will assemble after replication. If CID, the Drosophila CENP-A, is overexpressed and binds to a non-centromeric DNA sequence, a new, ectopic, centromere can be created at that site (Heun et al. 2006).
Other Possible Carriers of Epigenetic Information The many roles of non-coding RNA have become a focus of attention in recent years. Although there is still a lot to discover, some experiments show that RNA can be a carrier of epigenetic information through the germ line (Buckley et al. 2012). Small interfering RNAs (siRNAs) introduced into Caenorhabditis elegans can silence target genes for several generations.The mechanism involves proteins of the Argonaute family, which deliver the small RNA to the target nascent transcript, causing recruitment to the gene of silencing signals and H3K9 methylation (Burkhart et al. 2011). The persistence of silencing through several generations involves heritable expression of siRNA, and not merely serial dilution, and is dependent on expression of an Argonaute family gene, hrde-1 (Burton, Burkhart, and Kennedy 2011). Organisms lacking this gene are increasingly defective in gamete formation in successive generations, and after three to five generations are sterile. Thus the ability to transmit epigenetic information carried by RNA is an essential trait. RNA is also implicated in paramutation, studied extensively in maize, in which expression of one allele affects the other allele in a way that can be inherited for many generations, and in the absence of the first allele (Arteaga-Vazquez, Alberto, and Chandler 2010; Arteaga-Vazquez et al. 2010). Epigenetic effects have been well documented in vertebrates: a classical example is the Agouti mouse, in which coat color is determined by the DNA methylation state of an enhancer driving the Agouti gene (Dolenoy 2008). In wild type mice, this gene is expressed transiently during development and leads to the normal brown color. In the engineered mouse, a retrotransposon, introduced upstream, drives extended Agouti expression and a resulting yellow coat color. DNA methylation, which occurs with a certain frequency, can impair the enhancer activity of the retrotransposon, leading to coats of intermediate color.The coat color can be partially transmitted through the germ line, showing that the methylation pattern is partly preserved. The fact that DNA methylation can be affected by external signals (such as nutrition) means that the heritable phenotypes can be affected by environment. In other examples of epigenetic transmission, patterns of imprinting are transmitted through cell division: parent-of-origin silencing, often associated with allele-specific 142
Perspectives in Biology and Medicine
The Evolution of Epigenetics
DNA methylation patterns, has been observed for many gene systems in mouse and human, including the Igf2/H19 locus, Igf2r, and Kcnq1. In four cell embryos in mice, X chromosome inactivation is also restricted to the paternal allele. Other mechanisms independent of DNA sequence can also lead to phenotypic changes, though it is not clear whether they should be included in a definition of epigenetic phenomena. For example, the heat shock protein Hsp90 can act as a chaperone to maintain the correct folding and function of some target proteins carrying mutations that would otherwise lead to misfolding. However, if Hsp90 is mutated or suffers “overload,” novel phenotypes can arise from the now misfolded target proteins. It has been pointed out that under certain circumstances these phenotypes could make it easier for genetic selection to occur, fixing the trait in the genome (Rutherford and Lindquist 1998).
What Should the Definition of Epigenetics Be, and Does It Matter? The accepted definition of the term epigenetics has changed dramatically over the past 70 years, and this review has described some of the data that have been used to refine that definition. It should be obvious that any definition of this kind is an arbitrary choice, presumably designed to summarize a set of ideas succinctly. With the accumulation of vast amounts of information about genome organization, chromatin structure, and gene expression, arguments about definitions of epigenetics have become rather acrimonious. One obvious criticism is that many of what are claimed to be “epigenetic mechanisms” do not fit definition D2 described above: although they do not involve changes in DNA sequence, they are not mitotically or meiotically heritable. This can be confusing because, as discussed above, there is evidence in some cases for inheritance of histone modifications or histone variants, but in many (perhaps most) cases they are not inherited. Accordingly, there is a tendency recently to refer to these as “epigenetic marks,” in the sense that they may under some circumstances be part of an epigenetic apparatus. There is, however, a second layer of discontent with the idea that chromatin structure can convey epigenetic information. This has to do with the more fundamental question of whether chromatin structure can ever convey regulatory information in the absence of sequence-specific transcription factors. For that reason, as the result of a conference devoted to this question, a new definition (D3) was recently proposed: to qualify as an epigenetic process, an epigenetic state should be initiated by a specific process that then disappears, leaving the epigenetic mark to be propagated by a mechanism that replicates the mark through cell division and (in a few cases) in the germ line. As discussed above, there is no question that the initial signals carrying sequence-specific information have to come from transcription factors, but the remaining question is whether, once established, chromatin structure is sufficient to transmit this information.
winter 2014 • volume 57, number 1
143
Gary Felsenfeld
Figure 5 A typical proposed scheme showing how histone modifications could spread along chromatin (Bannister et al. 2001; Jenuwein 2001). Here, the same mechanism is invoked to suggest how histone methylation could propagate during DNA replication. A protein (Me recognition, solid fill) binds specifically to the methylated site (lysine 9) on histone H3 of the old nucleosomes (spheres, shaded fill) derived from the unreplicated DNA. This protein in turn recruits the methytransferase (Histone Me transferase, white fill) responsible for modifying the histone. The newly deposited nucleosomes (spheres, open fill) are modified by the methyltransferase associated with the adjacent old nucleosome. Evidence for such a mechanism is provided by the Polycomb system for propagation of histone H3 lysine 27 methylation (Hansen et al. 2008).
This is not a straightforward question. In the case of active genes, even when the “epigenetic” information carried by the chromatin structure is preserved through mitosis or meiosis, the presence of the active structure afterward will serve to facilitate recruitment of transcription factors. If, on the other hand, the chromatin structure is not preserved, then transcription factors will presumably function to reestablish it. In a cyclic system, cause and effect are in a sense interchangeable, and it is hard to construct a critical experiment. It is easier to see how chromatin structure could provide heritable silencing signals; mechanisms for transmission of such signals have been described above. Given that genes in the eukaryotic genome tend to be repressed until activated, rather than the reverse, this could represent an important function of chromatin, as suggested by Stedman and Stedman (1950) many years ago. The evolution of the definition of epigenetics shows how scientific definitions can change to accommodate increased knowledge. Chromatin was originally defined in terms of characteristic staining properties within the nucleus, and this definition 144
Perspectives in Biology and Medicine
The Evolution of Epigenetics
changed only after the molecular identity of its principal components had been established. But those two definitions describe the same material. In the case of epigenetics, the change between definitions D1 (strictly developmental) and D2 (molecular) was more extreme: the overlap between the processes described by the two definitions leaves large mutually excluded territories. Definition D3 seems to have been generated in part by a desire to make the definition D2 conform to the simple two-state mechanism initially described for the switch between the lytic and lysogenic states in lambda phage. It also focuses attention on early work by Weintraub, showing that somatic cells can be reprogrammed by expression of a single gene (MyoD), and by the more recent detailed understanding of the molecular changes associated with early differentiation of embryonic stem cells (ESCs), where altered expression of a small number of proteins can set in motion a series of “irreversible” changes in daughter cell gene expression patterns (Davis, Weintraub, and Lassar 1987; Yamanaka and Blau 2010). Such mechanisms can be described entirely in terms of changes in expression of transcription factors. Similarly, detailed analysis of regulatory gene expression patterns in developing sea urchin embryos has shown that they can be represented as a series of self-stabilizing stimulatory or inhibitory states that can be accounted for entirely in terms of binding of sequence-specific activators and repressors (Peter and Davidson 2011). Of course, this cannot mean that chromatin structure—histone modifications, nucleosome remodeling, the interrelated process of DNA methylation—and the vast array of enzymes that precisely control it are unimportant.These modifications are an essential part of the apparatus that carries out the plan initiated by the sequence-specific transcription factors. A vast literature describes major effects on phenotype that result from interference in chromatin modification systems. As discussed earlier, silencing chromatin structures can reduce expression of inactivated genes to levels not possible in prokaryotes. Chromatin structural changes can also serve as a “memory” to integrate multiple activation signals over time through changes in histone modifications and nucleosome positioning, so that not all sequence-specific transcription factors need be present simultaneously at a promoter or enhancer (Miranda et al. 2013;Voss et al. 2011). In part because the biochemical pathways involved appear to provide potential clinical targets, studies of “epigenetic” modifications of chromatin have attracted strong research funding. This seems entirely justified by the central role that chromatin plays in carrying out the instructions of sequence-specific transcription factors. Unfortunately, the controversy over the definition of epigenetics has become entangled with the question of whether research on chromatin should be supported—if in fact chromatin modifications don’t fit that definition. The evolution of the definition of epigenetics thus appears to have involved scientific, sociological, and economic influences. Meanwhile, with the recent, more detailed understanding of the ways in which the program of gene expression is controlled from embryonic stem cell to developing embryo, or in the developing
winter 2014 • volume 57, number 1
145
Gary Felsenfeld
sea urchin embryo, the original goal of Waddington seems to have been achieved. We do now understand how the DNA and its environment in the zygote interact ultimately to carry out the developmental plan.We also understand that this involves changing patterns of specific protein-DNA interactions, which serve to trigger complex biochemical reactions, including modifications of chromatin structure. As part of that program, the chromatin structural changes can in a few documented cases be passed on through mitosis or meiosis as a way to help reestablish in progeny a particular activity state of a given gene. The extent to which that mechanism is used is still under investigation, and it may be excessive to hold a definition hostage to the results of such studies.Thus, this might be an appropriate moment to return the term epigenetics to its original usage, which would encompass all of those mechanisms, and where it would call attention to how far we have come since 1942, when Waddington first defined it.
References Arteaga-Vazquez, Mario Alberto, and Vicki Lynn Chandler. 2010. “Paramutation in Maize: RNA Mediated Trans-generational Gene Silencing.” Curr Opin Genet Dev 20 (2): 156– 63. doi: 10.1016/j.gde.2010.01.008. Arteaga-Vazquez, Mario Alberto, et al. 2010. “RNA-Mediated Trans-communication Can Establish Paramutation at the b1 Locus in Maize.” Proc Natl Acad Sci USA 107 (29): 12986–91. doi: 10.1073/pnas.1007972107. Avery, Oswald T., Colin M. Macleod, and Maclyn McCarty. 1944. “Studies on the Chemical Nature of the Substance Inducing Transformation of Pneumococcal Types: Induction of Transformation by a Desoxyribonucleic Acid Fraction Isolated From Pneumococcus Type III.” J Exp Med 79 (2): 137–58. Bannister, Andrew J., et al. 2001. “Selective Recognition of Methylated Lysine 9 on Histone H3 by the HP1 Chromo Domain.” Nature 410 (6824): 120–24. doi: 10.1038/35065138. Beisel, Christian, and Renato Paro. 2011. “Silencing Chromatin: Comparing Modes and Mechanisms.” Nat Rev Genet 12 (2): 123–35. doi: 10.1038/nrg2932. Bird, Adrian P. 1978. “Use of Restriction Enzymes to Study Eukaryotic DNA Methylation: II. The Symmetry of Methylated Sites Supports Semi-conservative Copying of the Methylation Pattern.” J Mol Biol 118 (1): 49–60. Briggs, Robert, and Thomas J. King. 1952. “Transplantation of Living Nuclei from Blastula Cells into Enucleated Frogs’ Eggs.” Proc Natl Acad Sci USA 38 (5): 455–63. Brownell, James E., et al. 1996. “Tetrahymena Histone Acetyltransferase A: A Homolog to Yeast Gcn5p Linking Histone Acetylation to Gene Activation.” Cell 84 (6): 843–51. Buckley, Bethany A., et al. 2012.“A Nuclear Argonaute Promotes Multigenerational Epigenetic Inheritance and Germline Immortality.” Nature 489 (7416): 447–51. doi: 10.1038/ nature11352. Burkhart, Kirk B., et al. 2011. “A Pre-mRNA-associating Factor Links Endogenous SiRNAs to Chromatin Regulation.” PLoS Genet 7 (8): e1002249. doi: 10.1371/journal. pgen.1002249. Burton, Nick O., Kirk B. Burkhart, and Scott Kennedy. 2011. “Nuclear RNAi Maintains Heritable Gene Silencing in Caenorhabditis Elegans.” Proc Natl Acad Sci USA 108 (49): 19683–88. doi: 10.1073/pnas.1113310108. 146
Perspectives in Biology and Medicine
The Evolution of Epigenetics Davis, Robert L., Harold Weintraub, and Andrew B. Lassar. 1987. “Expression of a Single Transfected cDNA Converts Fibroblasts to Myoblasts.” Cell 51 (6): 987–1000. Gilbert, Scott F. 2012. Commentary: ‘The epigenotype’ by Conrad Hal Waddington. Int J Epidemiol 41 (1): 20–23. doi: 10.1093/ije/dyr186. Gilbert,Walter, and Benno Müller-Hill. 1966.“Isolation of the Lac Repressor.” Proc Natl Acad Sci USA 56 (6): 1891–98. Haig, David. 2004. “The (Dual) Origin of Epigenetics.” Cold Spring Harb Symp Quant Biol 69: 67–70. doi: 10.1101/sqb.2004.69.67. Hansen, Klaus H., et al. 2008. “A Model for Transmission of the H3K27me3 Epigenetic Mark.” Nat Cell Biol 10 (11): 1291–300. doi: 10.1038/ncb1787. Henikoff, Steven, and Ali Shilatifard. 2011. “Histone Modification: Cause or Cog?” Trends Genet 27 (10): 389–96. doi: 10.1016/j.tig.2011.06.006. Heun, Patrick, et al. 2006. “Mislocalization of the Drosophila Centromere-Specific Histone CID Promotes Formation of Functional Ectopic Kinetochores.” Dev Cell 10 (3): 303–15. doi: 10.1016/j.devcel.2006.01.014. Holliday, R., and J. E. Pugh. 1975. “DNA Modification Mechanisms and Gene Activity During Development.” Science 187 (4173): 226–32. Im, Hogune, et al. 2005. “Chromatin Domain Activation Via GATA-1 Utilization of a Small Subset of Dispersed GATA Motifs Within a Broad Chromosomal Region.” Proc Natl Acad Sci USA 102 (47): 17065–70. doi: 10.1073/pnas.0506164102. Jacob, François, and Jacques Monod. 1961.“Genetic Regulatory Mechanisms in the Synthesis of Proteins.” J Mol Biol 3: 318–56. Jenuwein, Thomas. 2001. “Re-SET-ting Heterochromatin by Histone Methyltransferases.” Trends Cell Biol 11 (6): 266–73. Jin, Chunyuan, and Gary Felsenfeld. 2007.“Nucleosome Stability Mediated by Histone Variants H3.3 and H2A.Z.” Genes Dev 21 (12): 1519–29. doi: 10.1101/gad.1547707. Kim, J. A., et al. 2012. “Mutagenesis of Pairwise Combinations of Histone Amino-Terminal Tails Reveals Functional Redundancy in Budding Yeast.” Proc Natl Acad Sci USA 109 (15): 5779–84. doi: 10.1073/pnas.1203453109. Kornberg, Roger D., and Jean O. Thomas. 1974. “Chromatin Structure: Oligomers of the Histones.” Science 184 (4139): 865–68. Laskey, R. A., and J. B. Gurdon. 1970. “Genetic Content of Adult Somatic Cells Tested by Nuclear Transplantation from Cultured Cells.” Nature 228 (5278): 1332–34. Luger, Karolin, et al. 1997. “Crystal Structure of the Nucleosome Core Particle at 2.8 A Resolution.” Nature 389 (6648): 251–60. doi: 10.1038/38444. Margueron, Raphaël, and Danny Reinberg. 2010. “Chromatin Structure and the Inheritance of Epigenetic Information.” Nat Rev Genet 11 (4): 285–96. doi: 10.1038/nrg2752. McKittrick, Erin, et al. 2004. “Histone H3.3 is Enriched in Covalent Modifications Associated with Active Chromatin.” Proc Natl Acad Sci USA 101 (6): 1525–30. doi: 10.1073/ pnas.0308092100. Miranda, Tina B., et al. 2013. “Reprogramming the Chromatin Landscape: Interplay of the Estrogen and Glucocorticoid Receptors at the Genomic Level.” Cancer Res 73 (16): 5130–39. doi: 10.1158/0008-5472.CAN-13-0742. Morgan, Thomas H. 1911. “An Attempt to Analyze the Constitution of the Chromosomes on the Basis of Sex-Linked Inheritance in Drosophila.” J Exp Zool 11 (4): 365–414. Muller, Hans J. 1930. “Types of Visible Variations Induced by X-rays in Drosophila.” J Genet 22: 299–334. winter 2014 • volume 57, number 1
147
Gary Felsenfeld Ng, Ray Kit, and J. B. Gurdon. 2008.“Epigenetic Memory of an Active Gene State Depends on Histone H3.3 Incorporation into Chromatin in the Absence of Transcription.” Nat Cell Biol 10 (1): 102–109. doi: 10.1038/ncb1674. Peter, Isabelle S., and Eric H. Davidson. 2011. “A Gene Regulatory Network Controlling the Embryonic Specification of Endoderm.” Nature 474 (7353): 635–39. doi: 10.1038/ nature10100. Ptashne, Mark. 1967. “Isolation of the Lambda Phage Repressor.” Proc Natl Acad Sci USA 57 (2): 306–13. Ptashne, Mark. 1992. A Genetic Switch: Phage l and Higher Organisms, 2nd ed. Malden: Blackwell Science. Ptashne, Mark, Alexander D. Johnson, and Carl O. Pabo. 1982. “A Genetic Switch in a Bacterial Virus.” Sci Am 247 (5): 128–30, 132, 134–40. Razin, Aharon, and Howard Cedar. 1977. “Distribution of 5-Methylcytosine in Chromatin.” Proc Natl Acad Sci USA 74 (7): 2725–28. Riggs, Arthur D. 1975. “X Inactivation, Differentiation, and DNA Methylation.” Cytogenet Cell Genet 14 (1): 9–25. Rutherford, Suzanne L., and Susan Lindquist. 1998. “Hsp90 as a Capacitor for Morphological Evolution.” Nature 396 (6709): 336–42. doi: 10.1038/24550. Simon, Jeffrey A., and Robert E. Kingston. 2013. “Occupying Chromatin: Polycomb Mechanisms for Getting to Genomic Targets, Stopping Transcriptional Traffic, and Staying Put.” Mol Cell 49 (5): 808–24. doi: 10.1016/j.molcel.2013.02.013. Stedman, Edgar, and Ellen Stedman. 1950. “Cell Specificity of Histones.” Nature 166 (4227): 780–81. Stein, Reuven, et al. 1982. “Clonal Inheritance of the Pattern of DNA Methylation in Mouse Cells.” Proc Natl Acad Sci USA 79 (1): 61–65. Sturtevant, A. H. 1925. “The Effects of Unequal Crossing Over at the Bar Locus in Drosophila.” Genetics 10 (2): 117–47. Taunton, Jack, Christian A. Hassig, and Stuart L. Schreiber. 1996. “A Mammalian Histone Deacetylase Related to the Yeast Transcriptional Regulator Rpd3p.” Science 272 (5260): 408–11. Voss,Ty C., et al. 2011.“Dynamic Exchange at Regulatory Elements During Chromatin Remodeling Underlies Assisted Loading Mechanism.” Cell 146 (4): 544–54. doi: 10.1016/j. cell.2011.07.006. Waddington, Conrad H. 1942. “The Epigenotype.” Endeavour 1: 18–20. Wilson, Edmund B. 1900. The Cell in Development and Inheritance, 2nd ed. New York: Macmillan. Yamanaka, Shinya, and Helen M. Blau. 2010. “Nuclear Reprogramming to a Pluripotent State by Three Approaches.” Nature 465 (7299): 704–12. doi: 10.1038/nature09229.
148
Perspectives in Biology and Medicine
Perspectives in Biology and Medicine, Volume 57, Number 1, Winter 2014, pp. 132-148 (Article) 3XEOLVKHGE\7KH-RKQV+RSNLQV8QLYHUVLW\3UHVV DOI: 10.1353/pbm.2014.0004
For additional information about this article http://muse.jhu.edu/journals/pbm/summary/v057/57.1.felsenfeld.html
Access provided by your local institution (24 Apr 2015 05:20 GMT)
The Evolution of Epigenetics
Gary Felsenfeld
ABSTRACT Early studies of the developing embryo raised the question of how
a fertilized egg could give rise to a complex multicellular organism containing many different kinds of cells. The term epigenetics originally referred to the study of these processes. With the advent of detailed knowledge of mechanisms of gene expression, this definition was superseded by another: epigenetics concerned the transmission of phenotype through mitosis or the germ line by mechanisms that did not involve changes in the DNA sequence. Much effort has been spent in attempting to identify and characterize these events. Work initially focused on DNA methylation as an epigenetic mark, but more recently there has been an emphasis on histone modifications as possible carriers of epigenetic information. However, there is confusion between situations in which the modifications may be propagated through cell division, thus helping to maintain a pattern of gene expression, and situations in which the modifications are simply part of the transcriptional apparatus. Arguments about the role of the histones have led to a reexamination of the definition of epigenetics and the primary events in development leading to cell type specific gene expression patterns.
S
ince the early days of embryology, a central puzzle for biologists has been how a fertilized egg can execute a clearly defined and reproducible program that leads ultimately to a complex organism. It was clear that all of the information
Laboratory of Molecular Biology, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, MD 20892. Email: [email protected]. This work was supported by the Intramural Research Program of the National Institute of Diabetes and Digestive and Kidney Diseases, NIH. Perspectives in Biology and Medicine, volume 57, number 1 (winter 2014): 132–148. © 2014 by Johns Hopkins University Press
132
The Evolution of Epigenetics
necessary to create the adult must already reside in the zygote, but how that information was translated into a complex organism was obscure. Even as recently as the late 1940s, the molecular mechanisms associated with early development were unknown and, in a sense, unknowable, given the available knowledge and technology. In the preceding centuries, embryologists had described in detail the progression of events during embryonic development, in both vertebrates and invertebrates. Two kinds of developmental models were considered. One assumed that the zygote already contained a preformed structure that simply expanded; the other proposed that development involved a preprogrammed series of chemical reactions required to generate the observed complexity. The discovery of chromosomes by Flemming in 1879 and the accumulating evidence that they carried the genetic information seemed to eliminate the first model. (For a complete description of the state of knowledge in 1900, see Wilson 1900.) In 1911, Morgan reported experiments in Drosophila that led him to two principal conclusions: “First, that sex-limited inheritance is explicable on the assumption that one of the material factors of a sex-limited character is carried by the same chromosomes that carry the material factor for femaleness. Second, that the ‘association’ of certain characters in inheritance is due to the proximity in the chromosomes of the chemical substances (factors) that are essential for the production of those characters” (365). Although it became accepted that the genetic information specifying the organism must reside in the chromosomes, and although individual “characters” could be associated with specific chromosomal loci, the chemical nature of the storage mechanism was unknown, as was the connection between the genes and the developmental process. At the same time, embryologists were cataloging the events of development using a different vocabulary and point of view. As described in a recent account of this period, embryologists were not especially interested in genes, because they were seen as static carriers of genetic information and therefore not capable of giving rise to a developmental program that varied among cell types (Gilbert 2012). Embryology and genetics were seen as distinct disciplines, addressing different questions about distinct mechanisms. In publications beginning in the late 1930s, embryologist C. H. Waddington attempted to make clear the connection between genes and developmental processes.The term epigenetics was used first by Waddington in 1942 to describe the study of the “causal mechanisms” by which “the genes of the genotype bring about phenotypic effects” (18). He saw clearly that specific patterns of expression of multiple genes were associated with each developmental stage and cell type. Waddington’s definition of epigenetics (which will be called D1) thus corresponded more or less to our contemporary view of the developmental process. All of these definitions were necessarily imprecise, because the roles of the many components of the chromosomes, as well as the other contents of the nucleus and cytoplasm, were not clear. Early work of A. H. Sturtevant (1925) and H. J. Muller (1930) had shown, however, that certain chromosome translocations in Drosophila
winter 2014 • volume 57, number 1
133
Gary Felsenfeld
could result in phenotypic changes even though there was no observable change in the chromosome structure within the displaced region. Many subsequent studies of such position effects made it clear that, whatever the molecular structure and mode of action of the individual “genes,” they did not function independently of a more extended environment. During this period, there was continuing debate about the relative contributions of the nucleus and cytoplasm to heredity (for a detailed description of these varied points of view, see review by Haig 2004). This debate extended not only beyond the 1944 experiments of Avery and colleagues, which provided strong evidence that DNA was the genetic material, but also beyond 1953, when the revelation of the DNA structure made clear where and how the genetic information might be stored.There was, in fact, continued doubt that DNA alone, free of proteins or other cellular components, could carry out this function. The idea that somatic cell nuclei contained the full complement of DNA was not confirmed until the experiments of Briggs and King (1952) and the definitive study by Laskey and Gurdon (1970), which showed that a nucleus from a somatic cell, introduced into an enucleated oocyte, could give rise to a normal adult. The problem of development, starting with the zygote, seemed to be reduced to a question of how individual genes were controlled in different cell types. Phenomena such as position effect variegation and X chromosome inactivation indicated, however, that heritable phenotypes might involve something in addition to DNA sequence. Early studies of gene regulation in bacteria and bacterial viruses suggested ways in which cellular expression states could be preserved through cell division. A trivial example would be a system in which a gene product was required for the activation of its own gene: provided that dilution during cell division did not reduce the abundance of the protein below some critical level, the active gene state would be maintained indefinitely through cell division. A more elaborate proposal from Delbrück involved a hypothetical pair of pathways, each producing an inhibitor of the other, and giving rise to a bi-stable expression system. An actual example of such a mechanism is provided by the switch between the lytic and lysogenic states in lambda bacteriophage, in which transient expression of a protein is sufficient to trigger a switch in states that is maintained through rounds of bacterial cell division (Ptashne 1992; Ptashne, Johnson, and Pabo 1982). In 1975, Riggs and Holliday and Pugh proposed that DNA methylation might in principle serve as a stable mark that could affect binding of transcription factors, hence gene expression, and that also could be propagated through cell division, if—as clearly stated by Riggs—there existed an enzyme that could methylate preferentially hemi-methylated sites, so that a palindromic site methylated on one strand would automatically be methylated on the second strand following replication (Figure 1). The existence of such systems was subsequently demonstrated by Bird (1978), and work in many laboratories has demonstrated the important role of DNA methylation in regulation of gene expression in vertebrates (Razin and Cedar 1977; Stein et al. 1982). 134
Perspectives in Biology and Medicine
The Evolution of Epigenetics
Figure 1 Propagation of DNA methylation patterns during DNA replication. A typical MeCpG site is modified on both strands. Following replication, the hemi-methylated site is recognized by the methyltransferase Dnmt1, which methylates the CpG on the newly synthesized strand.
The persistent debate about the relative contribution of mechanisms of cellular inheritance not connected directly to DNA sequence, probably reinforced by the increasingly detailed understanding of gene regulatory mechanisms, ultimately led to a change of focus in the way that the word epigenetics was used. The new definition (definition D2) became “the study of mitotically or meiotically heritable changes in gene function that cannot be explained by changes in DNA sequence.” To the extent that this might be viewed as a refinement of Waddington’s definition, it carries the unfortunate connotation that developmental processes should be ascribed principally to epigenetic mechanisms as newly defined—a point of view that has led to some confusion, as discussed below. DNA methylation was not the only phenomenon that could fit the new definition of epigenetics. In Drosophila, which lacks significant DNA methylation, the Polycomb protein complex was shown to shut down “permanently” genes no longer required in development, a state transmitted through cell division (for a recent review, see Simon and Kingston 2013). Specific X chromosome inactivation in vertebrate females appeared to be transmitted to daughter chromosomes, and at imprinted loci such as the Igf2/H19 region, only one of the two alleles was expressed, in a parent-of-origin–specific manner maintained through cell division. DNA methylation was implicated at least partially in these phenomena, but there winter 2014 • volume 57, number 1
135
Gary Felsenfeld
were other, major differences between the structures associated with the active and the inactive chromosome or alleles.
Epigenetics as Chromatin Structure The earliest experimental information about how gene expression was regulated came from studies in bacteria and bacteriophage. Jacob and Monod (1961) characterized the lac operon in E. coli; by the mid-1960s, the Gilbert laboratory had identified specific DNA-binding lac repressor protein (Gilbert and MüllerHill 1966), and Ptashne (1967) had isolated the lambda phage repressor.The model of transcriptional repression (and later, activation) through binding of regulatory proteins near the transcription start site, at “promoters,” was extended to eukaryotes, where it became clear that more distant sites, “enhancers,” also contributed to regulation of expression. It had long been obvious, however, that the organization of the genetic material in eukaryotes was much more elaborate: within the nucleus, the DNA is packaged as chromatin, a complex of DNA with proteins, principally the positively charged histones.These were at first envisioned as functioning to compact the large amounts of genomic DNA so that they would fit into the nucleus, and they do serve that purpose. It was also obvious that the histones might prevent DNA from being transcribed or replicated, unless there were specific mechanisms to modify or displace them. Well before the work in bacteria described above, and before it was proven that all somatic cells contained the same complement of DNA, Stedman and Stedman (1950) proposed that the histones acted as transcriptional repressors to silence genes in a cell-type specific manner, which would explain how different cell types had different phenotypes—the question that embryologists had been asking for many years. That point of view has in part survived, with modifications, in many models of chromatin function. One role of chromatin function is certainly to reduce expression of silenced genes to levels much lower than can be observed in prokaryote systems, but this model does not explain where the information comes from to instruct the histones which genes to repress, so the fundamental question is not answered. It was another quarter-century before it was discovered that the histones organized DNA through the formation of repeating subunits, called nucleosomes (Kornberg and Thomas 1974). Each nucleosome contains an octamer of the “core” histones: two each of H2A, H2B, H3, and H4, with two superhelical turns, 165 base pairs of DNA, wrapped around the outside, providing the first level of DNA compaction inside the nucleus (Figure 2). A single molecule of a separate histone family (H1) is typically bound in vivo at the site where the DNA enters and exits the nucleosome. Most work, however, has been done with so-called nucleosome core particles, which have 147 base pairs of DNA wrapped in 1¾ turns, and lack H1. The fact that the core histones are highly conserved in evolution suggested that they must have functions besides DNA compaction. We now know, through work 136
Perspectives in Biology and Medicine
The Evolution of Epigenetics
Figure 2 High resolution structure of the nucleosome core particle. The alpha-helical domains of the four core histones (two copies each of H2A, H2B, H3, and H4) are visible. The histone “tails,” which extend from the particle, are accessible for modification. Source: Luger et al. 1997.
in many laboratories, that the core histones are targets of covalent modifications that modify their properties and can to a considerable extent be correlated with the function of the DNA sequences with which they are associated. Most but not all of these modifications are located on the amino terminal “tails” of the histones, which protrude from the nucleosome and are accessible to enzymes (see Figure 2). Some of the known modifications are shown in Figure 3. They include acetylation of lysines, methylation of lysines and arginines, and phosphorylation of serines and threonines. This is further complicated by the ability of lysine and arginine residues to be either mono-, di-, or (in the case of lysine) tri-methylated. Each modification typically requires a specific enzyme, but there may be more than one specific enzyme capable of that modification. Which one of these is used at a particular site in vivo may depend on the nature of the site. All of these modifications are reversible: again, specific enzymes are responsible for removing each kind of modification. Do these modifications affect function, and if so, is there a simple relationship between a particular modification and a corresponding function? Early experiments, especially in budding yeast (S. cerevisiae), showed that interfering with the function of individual histone modifying enzymes led to distinct phenotypes (Brownell et al. 1996; Taunton, Hassig, and Schreiber 1996). This has been confirmed by countless experiments in cells from higher eukaryotes. Thus, there can be little question that these histone marks have essential functions. But what are these functions? It
winter 2014 • volume 57, number 1
137
Gary Felsenfeld
Figure 3 Some of the chemical modifications found on the amino terminal tails of histones H3 and H4, and the associations often found between certain modifications and transcriptional activity.
was originally suggested that each modification would have a specific, identifiable effect on some aspect of DNA biochemistry, constituting a code that would allow correlating the modification with a particular DNA-dependent process: transcription initiation or elongation, enhancer or promoter activity, DNA repair or DNA replication. With the advent of ChIP-seq methods for mapping individual histone modifications genome-wide, it became possible to address these questions. It was evident that certain modifications had a high probability of being associated with particular regulatory features: for example, transcriptionally active genes are typically marked with acetylated histones.There are also strong correlations of certain histone H3 methylation marks with enhancers and promoters. Other methylation marks in higher eukaryotes are associated with silenced heterochromatic regions. Certain sites may display specific modification patterns, perhaps cell cycle dependent. There is, however, no simple set of rules that can account for all of the varied combinations of observed modifications; unlike the genetic code, these patterns cannot be decoded unambiguously. In budding yeast, for example, it has recently been shown that acetylated lysines on the N terminal tails of histones H2A and H4 have a redundant effect on phenotype (Kim et al. 2012). Despite the complexity of these relationships, it is clear that there are important correlations between histone modifications and gene activity, and that interfering with the modification pathways can have major effects in vivo. For that reason, investigators began referring to the modifications as “epigenetic marks,” perhaps by analogy to the DNA methylation mark. This has given rise to a heated debate, 138
Perspectives in Biology and Medicine
The Evolution of Epigenetics
discussed further below, that has resulted in attempts to further constrain the definition of epigenetics. But even without resorting to redefinition, the question remains of whether histone modifications, though clearly distinct from changes in DNA, can induce mitotically or meiotically heritable changes in gene function. To answer that question it would be helpful if we understood the biochemical relationship between histone modification and gene expression. Unfortunately, the answer appears to be getting more obscure as new data and points of view accumulate. Recent results, such as those of Kim et al. (2012), lend support to the idea that histone acetylation is generated as a consequence of transcription and serves to reduce the affinity of the histone octamer for DNA, making subsequent transcription events easier (Henikoff and Shilatifard 2011). Amusingly, this is an idea that was first advanced in the early studies of this modification, and there were in fact numerous reports showing that octamers containing acetylated histones are not bound as tightly to nucleosomal DNA. Even if one accepts this view that some modifications follow rather than precede transcription, it is important to remember that they do play an important role in transcriptional mechanisms; interference with the modification process has multiple effects on phenotype. Many proteins can recognize histone modifications as a signal to bind either more or less tightly to the nucleosome. For example, the protein HP1, often associated with silenced chromatin, binds to methylated lysine 9 of histone H3 (H3K9Me2/3), which in turn can lead to recruitment of the enzyme that carries out that methylation. Histone modifications can also lead to the recruitment of ATP-dependent nucleosome remodeling enzymes, capable of moving nucleosomes in such a way as to make specific DNA sequences more or less accessible. Some studies have documented an orderly series of events that occur during activation of specific genes (Im et al. 2005; Miranda et al. 2013). Furthermore, there can be a series of well-defined steps in which sites of DNA methylation can recruit enzymes that introduce “silencing” modifications, and in turn some sites of modification can recruit DNA methytransferase. Histone variants also contribute to regulation of gene expression. Among these, the H3 variant histone H3.3 confers reduced stability on nucleosomes in which it is incorporated; nucleosomes containing the variant H2A.Z in combination with H3.3 are even more unstable (Jin and Felsenfeld 2007). The H3 variant CENP-A is essential to centromere identity. Some of these variants have clear epigenetic functions, as discussed below.
Is Chromatin Structure Transmissible? To qualify as an epigenetic mechanism according to definition D2, a chromatin modification pattern or structure would have to be transmissible through cell division. The difficulty in addressing this question is in part related to the problem of distinguishing cause from effect. Can histone modifications, in the absence of other signals, create a specific active or silenced chromatin domain, or are they winter 2014 • volume 57, number 1
139
Gary Felsenfeld
always downstream consequences of earlier processes? It seems clear that during development the initial information to determine the activity state of a gene must come from sequence-specific transcription factors that recognize the regulatory elements associated with the gene. This is the only kind of interaction we know of that can convey that information (Figure 4). It follows that whatever the associated histone modifications, they must occur downstream of that initial event. If we are considering an individual cell, then the histone modification changes that follow binding of the transcription factor could most reasonably be viewed simply as biochemical processes that are part of the total regulatory apparatus: essential, but with no epigenetic connotations. As critics have often stated, it can be misleading to refer to histone modifications under these circumstances as “epigenetic marks.” The more relevant question, however, is whether histone marks, once established, can be transmitted through cell division or the germ line in such a way as to preserve the activity state of the associated genes without going through further de novo initiation by transcription factors as described above. In this model, chromatin structure, surviving through cell division, preserves the earlier-established “on” or “off ” state of the locus. Such behavior would qualify as epigenetic according to definition D2. One class of proposed mechanisms began with the observation that HP1 binds to H3K9Me2/3, and that HP1 in turn recruits the enzyme that carries out H3K9 methylation, creating in principle a system for propagation of the mark (Bannister et al. 2001; Jenuwein 2001). The functioning of such systems in vivo has been demonstrated most clearly for the Polycomb complex PRC2 and its associated modification, trimethylation of H3 lysine 27 (H3K27me3) (Hansen et al. 2008). PRC2 binds to H3K27me3, and binding to a methylated site helps activate the PRC2 complex member responsible for methylating H3K27, in principle allowing it to methylate a site on an adjacent nucleosome (Figure 5). A form of PRC2, in which one component was fused to GAL4, was targeted in cells to GAL4 binding sites upstream of a reporter gene integrated into the genome. This resulted in silencing of the reporter and increased H3K27 trimethylation. Most important, the methylation mark, which extended downstream of the target site for at least 1,600 base pairs, was maintained, due to the action of the endogenous PRC2 complex, over several population doublings after expression of GAL4-PRC2 was shut down. These and other results show how the Polycomb complex could propagate the K27 methylation mark along a chain of nucleosomes during interphase (Beisel and Paro 2011; Margueron and Reinberg 2010); the same mechanism could explain how the same signal could be maintained during and after cell division. Recent studies have shown that such processes are likely to be rather more complicated, perhaps involving another Polycomb complex, PRC1, or maintenance of Polycomb binding during DNA replication (see review by Simon and Kingston 2013). The evidence does seem to support some kind of epigenetic mechanism, but it is still not clear what biochemical processes, leading to silencing, are signaled by H3K27 trimethylation.
140
Perspectives in Biology and Medicine
The Evolution of Epigenetics
Figure 4 A simplified description of steps that may be associated with activation of gene transcription. The important point is that the initial step (top) must involve recognition of DNA sites (nucleotide sequences) near the gene by one or more specific regulatory proteins that target that gene. That is the signal that allows the remaining steps (below) to prepare the chromatin to accommodate RNA polymerase and transcription.
Histone variants have also been implicated in epigenetic processes. Because nucleosomes containing H3.3 are less stable than those containing the major variants and are found at transcriptionally active genes, it has been suggested that DNA covered with such H3.3 nucleosomes would be more readily transcribed (McKittrick et al. 2004; Ng and Gurdon 2008). Furthermore, nucleosomes containing H3.3 can be deposited throughout the cell cycle, whereas H3.1 or H3.2 are laid down only during S phase. Because the passage of RNA pol II tends to displace nucleosomes, the newly deposited ones will contain H3.3 and make transcription even easier. Most important, after replication the daughter gene copies will still contain H3.3 and tend to retain the “active” conformation, thus propagating that state. Experiments in which Xenopus somatic cell nuclei are transplanted into enucleated oocytes support a role for H3.3 in epigenetic transmission of phenotype (Ng and Gurdon 2008). A nucleus is taken from an embryonic endodermal cell, and the resulting embryo inappropriately expresses endoderm-specific genes in cells of the animal pole. If the experiment is repeated with a nucleus from one of these cells, the resulting embryo again has incorrectly programmed animal pole cells. These epigenetic effects are modulated by H3.3: overexpression of H3.3 increases and depletion decreases the number of inappropriately expressing cells.
winter 2014 • volume 57, number 1
141
Gary Felsenfeld
The most unambiguous example of an epigenetic process mediated solely by a histone is provided by the family of centromeric H3 variants, called CENP-A in vertebrates. Nucleosomes containing CENP-A are found only at centromeres, and they are essential to centromeric function during mitosis. It has been shown in Drosophila, and more recently in vertebrates, that the location of a centromere depends entirely on the presence of CENP-A. Although normal centromeres do contain special sequences, it is the presence of CENP-A that signals where centromeres will assemble after replication. If CID, the Drosophila CENP-A, is overexpressed and binds to a non-centromeric DNA sequence, a new, ectopic, centromere can be created at that site (Heun et al. 2006).
Other Possible Carriers of Epigenetic Information The many roles of non-coding RNA have become a focus of attention in recent years. Although there is still a lot to discover, some experiments show that RNA can be a carrier of epigenetic information through the germ line (Buckley et al. 2012). Small interfering RNAs (siRNAs) introduced into Caenorhabditis elegans can silence target genes for several generations.The mechanism involves proteins of the Argonaute family, which deliver the small RNA to the target nascent transcript, causing recruitment to the gene of silencing signals and H3K9 methylation (Burkhart et al. 2011). The persistence of silencing through several generations involves heritable expression of siRNA, and not merely serial dilution, and is dependent on expression of an Argonaute family gene, hrde-1 (Burton, Burkhart, and Kennedy 2011). Organisms lacking this gene are increasingly defective in gamete formation in successive generations, and after three to five generations are sterile. Thus the ability to transmit epigenetic information carried by RNA is an essential trait. RNA is also implicated in paramutation, studied extensively in maize, in which expression of one allele affects the other allele in a way that can be inherited for many generations, and in the absence of the first allele (Arteaga-Vazquez, Alberto, and Chandler 2010; Arteaga-Vazquez et al. 2010). Epigenetic effects have been well documented in vertebrates: a classical example is the Agouti mouse, in which coat color is determined by the DNA methylation state of an enhancer driving the Agouti gene (Dolenoy 2008). In wild type mice, this gene is expressed transiently during development and leads to the normal brown color. In the engineered mouse, a retrotransposon, introduced upstream, drives extended Agouti expression and a resulting yellow coat color. DNA methylation, which occurs with a certain frequency, can impair the enhancer activity of the retrotransposon, leading to coats of intermediate color.The coat color can be partially transmitted through the germ line, showing that the methylation pattern is partly preserved. The fact that DNA methylation can be affected by external signals (such as nutrition) means that the heritable phenotypes can be affected by environment. In other examples of epigenetic transmission, patterns of imprinting are transmitted through cell division: parent-of-origin silencing, often associated with allele-specific 142
Perspectives in Biology and Medicine
The Evolution of Epigenetics
DNA methylation patterns, has been observed for many gene systems in mouse and human, including the Igf2/H19 locus, Igf2r, and Kcnq1. In four cell embryos in mice, X chromosome inactivation is also restricted to the paternal allele. Other mechanisms independent of DNA sequence can also lead to phenotypic changes, though it is not clear whether they should be included in a definition of epigenetic phenomena. For example, the heat shock protein Hsp90 can act as a chaperone to maintain the correct folding and function of some target proteins carrying mutations that would otherwise lead to misfolding. However, if Hsp90 is mutated or suffers “overload,” novel phenotypes can arise from the now misfolded target proteins. It has been pointed out that under certain circumstances these phenotypes could make it easier for genetic selection to occur, fixing the trait in the genome (Rutherford and Lindquist 1998).
What Should the Definition of Epigenetics Be, and Does It Matter? The accepted definition of the term epigenetics has changed dramatically over the past 70 years, and this review has described some of the data that have been used to refine that definition. It should be obvious that any definition of this kind is an arbitrary choice, presumably designed to summarize a set of ideas succinctly. With the accumulation of vast amounts of information about genome organization, chromatin structure, and gene expression, arguments about definitions of epigenetics have become rather acrimonious. One obvious criticism is that many of what are claimed to be “epigenetic mechanisms” do not fit definition D2 described above: although they do not involve changes in DNA sequence, they are not mitotically or meiotically heritable. This can be confusing because, as discussed above, there is evidence in some cases for inheritance of histone modifications or histone variants, but in many (perhaps most) cases they are not inherited. Accordingly, there is a tendency recently to refer to these as “epigenetic marks,” in the sense that they may under some circumstances be part of an epigenetic apparatus. There is, however, a second layer of discontent with the idea that chromatin structure can convey epigenetic information. This has to do with the more fundamental question of whether chromatin structure can ever convey regulatory information in the absence of sequence-specific transcription factors. For that reason, as the result of a conference devoted to this question, a new definition (D3) was recently proposed: to qualify as an epigenetic process, an epigenetic state should be initiated by a specific process that then disappears, leaving the epigenetic mark to be propagated by a mechanism that replicates the mark through cell division and (in a few cases) in the germ line. As discussed above, there is no question that the initial signals carrying sequence-specific information have to come from transcription factors, but the remaining question is whether, once established, chromatin structure is sufficient to transmit this information.
winter 2014 • volume 57, number 1
143
Gary Felsenfeld
Figure 5 A typical proposed scheme showing how histone modifications could spread along chromatin (Bannister et al. 2001; Jenuwein 2001). Here, the same mechanism is invoked to suggest how histone methylation could propagate during DNA replication. A protein (Me recognition, solid fill) binds specifically to the methylated site (lysine 9) on histone H3 of the old nucleosomes (spheres, shaded fill) derived from the unreplicated DNA. This protein in turn recruits the methytransferase (Histone Me transferase, white fill) responsible for modifying the histone. The newly deposited nucleosomes (spheres, open fill) are modified by the methyltransferase associated with the adjacent old nucleosome. Evidence for such a mechanism is provided by the Polycomb system for propagation of histone H3 lysine 27 methylation (Hansen et al. 2008).
This is not a straightforward question. In the case of active genes, even when the “epigenetic” information carried by the chromatin structure is preserved through mitosis or meiosis, the presence of the active structure afterward will serve to facilitate recruitment of transcription factors. If, on the other hand, the chromatin structure is not preserved, then transcription factors will presumably function to reestablish it. In a cyclic system, cause and effect are in a sense interchangeable, and it is hard to construct a critical experiment. It is easier to see how chromatin structure could provide heritable silencing signals; mechanisms for transmission of such signals have been described above. Given that genes in the eukaryotic genome tend to be repressed until activated, rather than the reverse, this could represent an important function of chromatin, as suggested by Stedman and Stedman (1950) many years ago. The evolution of the definition of epigenetics shows how scientific definitions can change to accommodate increased knowledge. Chromatin was originally defined in terms of characteristic staining properties within the nucleus, and this definition 144
Perspectives in Biology and Medicine
The Evolution of Epigenetics
changed only after the molecular identity of its principal components had been established. But those two definitions describe the same material. In the case of epigenetics, the change between definitions D1 (strictly developmental) and D2 (molecular) was more extreme: the overlap between the processes described by the two definitions leaves large mutually excluded territories. Definition D3 seems to have been generated in part by a desire to make the definition D2 conform to the simple two-state mechanism initially described for the switch between the lytic and lysogenic states in lambda phage. It also focuses attention on early work by Weintraub, showing that somatic cells can be reprogrammed by expression of a single gene (MyoD), and by the more recent detailed understanding of the molecular changes associated with early differentiation of embryonic stem cells (ESCs), where altered expression of a small number of proteins can set in motion a series of “irreversible” changes in daughter cell gene expression patterns (Davis, Weintraub, and Lassar 1987; Yamanaka and Blau 2010). Such mechanisms can be described entirely in terms of changes in expression of transcription factors. Similarly, detailed analysis of regulatory gene expression patterns in developing sea urchin embryos has shown that they can be represented as a series of self-stabilizing stimulatory or inhibitory states that can be accounted for entirely in terms of binding of sequence-specific activators and repressors (Peter and Davidson 2011). Of course, this cannot mean that chromatin structure—histone modifications, nucleosome remodeling, the interrelated process of DNA methylation—and the vast array of enzymes that precisely control it are unimportant.These modifications are an essential part of the apparatus that carries out the plan initiated by the sequence-specific transcription factors. A vast literature describes major effects on phenotype that result from interference in chromatin modification systems. As discussed earlier, silencing chromatin structures can reduce expression of inactivated genes to levels not possible in prokaryotes. Chromatin structural changes can also serve as a “memory” to integrate multiple activation signals over time through changes in histone modifications and nucleosome positioning, so that not all sequence-specific transcription factors need be present simultaneously at a promoter or enhancer (Miranda et al. 2013;Voss et al. 2011). In part because the biochemical pathways involved appear to provide potential clinical targets, studies of “epigenetic” modifications of chromatin have attracted strong research funding. This seems entirely justified by the central role that chromatin plays in carrying out the instructions of sequence-specific transcription factors. Unfortunately, the controversy over the definition of epigenetics has become entangled with the question of whether research on chromatin should be supported—if in fact chromatin modifications don’t fit that definition. The evolution of the definition of epigenetics thus appears to have involved scientific, sociological, and economic influences. Meanwhile, with the recent, more detailed understanding of the ways in which the program of gene expression is controlled from embryonic stem cell to developing embryo, or in the developing
winter 2014 • volume 57, number 1
145
Gary Felsenfeld
sea urchin embryo, the original goal of Waddington seems to have been achieved. We do now understand how the DNA and its environment in the zygote interact ultimately to carry out the developmental plan.We also understand that this involves changing patterns of specific protein-DNA interactions, which serve to trigger complex biochemical reactions, including modifications of chromatin structure. As part of that program, the chromatin structural changes can in a few documented cases be passed on through mitosis or meiosis as a way to help reestablish in progeny a particular activity state of a given gene. The extent to which that mechanism is used is still under investigation, and it may be excessive to hold a definition hostage to the results of such studies.Thus, this might be an appropriate moment to return the term epigenetics to its original usage, which would encompass all of those mechanisms, and where it would call attention to how far we have come since 1942, when Waddington first defined it.
References Arteaga-Vazquez, Mario Alberto, and Vicki Lynn Chandler. 2010. “Paramutation in Maize: RNA Mediated Trans-generational Gene Silencing.” Curr Opin Genet Dev 20 (2): 156– 63. doi: 10.1016/j.gde.2010.01.008. Arteaga-Vazquez, Mario Alberto, et al. 2010. “RNA-Mediated Trans-communication Can Establish Paramutation at the b1 Locus in Maize.” Proc Natl Acad Sci USA 107 (29): 12986–91. doi: 10.1073/pnas.1007972107. Avery, Oswald T., Colin M. Macleod, and Maclyn McCarty. 1944. “Studies on the Chemical Nature of the Substance Inducing Transformation of Pneumococcal Types: Induction of Transformation by a Desoxyribonucleic Acid Fraction Isolated From Pneumococcus Type III.” J Exp Med 79 (2): 137–58. Bannister, Andrew J., et al. 2001. “Selective Recognition of Methylated Lysine 9 on Histone H3 by the HP1 Chromo Domain.” Nature 410 (6824): 120–24. doi: 10.1038/35065138. Beisel, Christian, and Renato Paro. 2011. “Silencing Chromatin: Comparing Modes and Mechanisms.” Nat Rev Genet 12 (2): 123–35. doi: 10.1038/nrg2932. Bird, Adrian P. 1978. “Use of Restriction Enzymes to Study Eukaryotic DNA Methylation: II. The Symmetry of Methylated Sites Supports Semi-conservative Copying of the Methylation Pattern.” J Mol Biol 118 (1): 49–60. Briggs, Robert, and Thomas J. King. 1952. “Transplantation of Living Nuclei from Blastula Cells into Enucleated Frogs’ Eggs.” Proc Natl Acad Sci USA 38 (5): 455–63. Brownell, James E., et al. 1996. “Tetrahymena Histone Acetyltransferase A: A Homolog to Yeast Gcn5p Linking Histone Acetylation to Gene Activation.” Cell 84 (6): 843–51. Buckley, Bethany A., et al. 2012.“A Nuclear Argonaute Promotes Multigenerational Epigenetic Inheritance and Germline Immortality.” Nature 489 (7416): 447–51. doi: 10.1038/ nature11352. Burkhart, Kirk B., et al. 2011. “A Pre-mRNA-associating Factor Links Endogenous SiRNAs to Chromatin Regulation.” PLoS Genet 7 (8): e1002249. doi: 10.1371/journal. pgen.1002249. Burton, Nick O., Kirk B. Burkhart, and Scott Kennedy. 2011. “Nuclear RNAi Maintains Heritable Gene Silencing in Caenorhabditis Elegans.” Proc Natl Acad Sci USA 108 (49): 19683–88. doi: 10.1073/pnas.1113310108. 146
Perspectives in Biology and Medicine
The Evolution of Epigenetics Davis, Robert L., Harold Weintraub, and Andrew B. Lassar. 1987. “Expression of a Single Transfected cDNA Converts Fibroblasts to Myoblasts.” Cell 51 (6): 987–1000. Gilbert, Scott F. 2012. Commentary: ‘The epigenotype’ by Conrad Hal Waddington. Int J Epidemiol 41 (1): 20–23. doi: 10.1093/ije/dyr186. Gilbert,Walter, and Benno Müller-Hill. 1966.“Isolation of the Lac Repressor.” Proc Natl Acad Sci USA 56 (6): 1891–98. Haig, David. 2004. “The (Dual) Origin of Epigenetics.” Cold Spring Harb Symp Quant Biol 69: 67–70. doi: 10.1101/sqb.2004.69.67. Hansen, Klaus H., et al. 2008. “A Model for Transmission of the H3K27me3 Epigenetic Mark.” Nat Cell Biol 10 (11): 1291–300. doi: 10.1038/ncb1787. Henikoff, Steven, and Ali Shilatifard. 2011. “Histone Modification: Cause or Cog?” Trends Genet 27 (10): 389–96. doi: 10.1016/j.tig.2011.06.006. Heun, Patrick, et al. 2006. “Mislocalization of the Drosophila Centromere-Specific Histone CID Promotes Formation of Functional Ectopic Kinetochores.” Dev Cell 10 (3): 303–15. doi: 10.1016/j.devcel.2006.01.014. Holliday, R., and J. E. Pugh. 1975. “DNA Modification Mechanisms and Gene Activity During Development.” Science 187 (4173): 226–32. Im, Hogune, et al. 2005. “Chromatin Domain Activation Via GATA-1 Utilization of a Small Subset of Dispersed GATA Motifs Within a Broad Chromosomal Region.” Proc Natl Acad Sci USA 102 (47): 17065–70. doi: 10.1073/pnas.0506164102. Jacob, François, and Jacques Monod. 1961.“Genetic Regulatory Mechanisms in the Synthesis of Proteins.” J Mol Biol 3: 318–56. Jenuwein, Thomas. 2001. “Re-SET-ting Heterochromatin by Histone Methyltransferases.” Trends Cell Biol 11 (6): 266–73. Jin, Chunyuan, and Gary Felsenfeld. 2007.“Nucleosome Stability Mediated by Histone Variants H3.3 and H2A.Z.” Genes Dev 21 (12): 1519–29. doi: 10.1101/gad.1547707. Kim, J. A., et al. 2012. “Mutagenesis of Pairwise Combinations of Histone Amino-Terminal Tails Reveals Functional Redundancy in Budding Yeast.” Proc Natl Acad Sci USA 109 (15): 5779–84. doi: 10.1073/pnas.1203453109. Kornberg, Roger D., and Jean O. Thomas. 1974. “Chromatin Structure: Oligomers of the Histones.” Science 184 (4139): 865–68. Laskey, R. A., and J. B. Gurdon. 1970. “Genetic Content of Adult Somatic Cells Tested by Nuclear Transplantation from Cultured Cells.” Nature 228 (5278): 1332–34. Luger, Karolin, et al. 1997. “Crystal Structure of the Nucleosome Core Particle at 2.8 A Resolution.” Nature 389 (6648): 251–60. doi: 10.1038/38444. Margueron, Raphaël, and Danny Reinberg. 2010. “Chromatin Structure and the Inheritance of Epigenetic Information.” Nat Rev Genet 11 (4): 285–96. doi: 10.1038/nrg2752. McKittrick, Erin, et al. 2004. “Histone H3.3 is Enriched in Covalent Modifications Associated with Active Chromatin.” Proc Natl Acad Sci USA 101 (6): 1525–30. doi: 10.1073/ pnas.0308092100. Miranda, Tina B., et al. 2013. “Reprogramming the Chromatin Landscape: Interplay of the Estrogen and Glucocorticoid Receptors at the Genomic Level.” Cancer Res 73 (16): 5130–39. doi: 10.1158/0008-5472.CAN-13-0742. Morgan, Thomas H. 1911. “An Attempt to Analyze the Constitution of the Chromosomes on the Basis of Sex-Linked Inheritance in Drosophila.” J Exp Zool 11 (4): 365–414. Muller, Hans J. 1930. “Types of Visible Variations Induced by X-rays in Drosophila.” J Genet 22: 299–334. winter 2014 • volume 57, number 1
147
Gary Felsenfeld Ng, Ray Kit, and J. B. Gurdon. 2008.“Epigenetic Memory of an Active Gene State Depends on Histone H3.3 Incorporation into Chromatin in the Absence of Transcription.” Nat Cell Biol 10 (1): 102–109. doi: 10.1038/ncb1674. Peter, Isabelle S., and Eric H. Davidson. 2011. “A Gene Regulatory Network Controlling the Embryonic Specification of Endoderm.” Nature 474 (7353): 635–39. doi: 10.1038/ nature10100. Ptashne, Mark. 1967. “Isolation of the Lambda Phage Repressor.” Proc Natl Acad Sci USA 57 (2): 306–13. Ptashne, Mark. 1992. A Genetic Switch: Phage l and Higher Organisms, 2nd ed. Malden: Blackwell Science. Ptashne, Mark, Alexander D. Johnson, and Carl O. Pabo. 1982. “A Genetic Switch in a Bacterial Virus.” Sci Am 247 (5): 128–30, 132, 134–40. Razin, Aharon, and Howard Cedar. 1977. “Distribution of 5-Methylcytosine in Chromatin.” Proc Natl Acad Sci USA 74 (7): 2725–28. Riggs, Arthur D. 1975. “X Inactivation, Differentiation, and DNA Methylation.” Cytogenet Cell Genet 14 (1): 9–25. Rutherford, Suzanne L., and Susan Lindquist. 1998. “Hsp90 as a Capacitor for Morphological Evolution.” Nature 396 (6709): 336–42. doi: 10.1038/24550. Simon, Jeffrey A., and Robert E. Kingston. 2013. “Occupying Chromatin: Polycomb Mechanisms for Getting to Genomic Targets, Stopping Transcriptional Traffic, and Staying Put.” Mol Cell 49 (5): 808–24. doi: 10.1016/j.molcel.2013.02.013. Stedman, Edgar, and Ellen Stedman. 1950. “Cell Specificity of Histones.” Nature 166 (4227): 780–81. Stein, Reuven, et al. 1982. “Clonal Inheritance of the Pattern of DNA Methylation in Mouse Cells.” Proc Natl Acad Sci USA 79 (1): 61–65. Sturtevant, A. H. 1925. “The Effects of Unequal Crossing Over at the Bar Locus in Drosophila.” Genetics 10 (2): 117–47. Taunton, Jack, Christian A. Hassig, and Stuart L. Schreiber. 1996. “A Mammalian Histone Deacetylase Related to the Yeast Transcriptional Regulator Rpd3p.” Science 272 (5260): 408–11. Voss,Ty C., et al. 2011.“Dynamic Exchange at Regulatory Elements During Chromatin Remodeling Underlies Assisted Loading Mechanism.” Cell 146 (4): 544–54. doi: 10.1016/j. cell.2011.07.006. Waddington, Conrad H. 1942. “The Epigenotype.” Endeavour 1: 18–20. Wilson, Edmund B. 1900. The Cell in Development and Inheritance, 2nd ed. New York: Macmillan. Yamanaka, Shinya, and Helen M. Blau. 2010. “Nuclear Reprogramming to a Pluripotent State by Three Approaches.” Nature 465 (7299): 704–12. doi: 10.1038/nature09229.
148
Perspectives in Biology and Medicine
